\begin{table}[htbp]
\begin{center}
\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
\caption{Law of motion for the U.S. split for above and below the $30^\text{th}$ percentile of lag relative sales and negative of lag debt-to-asset ratio}
\begin{tabular}{l*{2}{c}}
\hline\hline
          &\multicolumn{1}{c}{(1)}&\multicolumn{1}{c}{(2)}\\
          &\multicolumn{1}{c}{$\text{log(M}_{i,t})$}&\multicolumn{1}{c}{$\text{log(M}_{i,t})$}\\
\hline
$\Delta \text{log(sales}_{i,t+1})$&    0.105         &    0.220\sym{**} \\
          &  (0.071)         &  (0.088)         \\
[1em]
$\text{log(M}_{i,t+1})$&    1.104\sym{***}&    1.037\sym{***}\\
          &  (0.008)         &  (0.006)         \\
\hline
\(R^{2}\) &    0.484         &    0.699         \\
Above $30^\text{th}$ percentile&       No         &      Yes         \\
Industry FEs&      Yes         &      Yes         \\
Year FEs  &      Yes         &      Yes         \\
Market share&    27.88         &    72.11         \\
\(N\)     &    71999         &    73701         \\
\hline\hline
\end{tabular}
\end{center}
\footnotesize{$*p < 0.10, ** p < 0.05, *** p < 0.01$; standard errors are clustered at the firm-level. We report the GMM estimator of the effects of $\Delta$ log($\text{sales}_{i,t+1}$) and log($\text{markup}_{i,t+1}$) on log($\text{markup}_{i,t}$) using four lags of log($\text{sales}_{i,t}$) and log($\text{markup}_{i,t}$) as instruments. We also include year and industry (SIC 2-digit codes) fixed effects. The second column reports the results for observations above the $30^\text{th}$ percentile of lag relative sales and negative of lag debt-to-asset ratio, while the first column reports the results for all other observations. We define relative sales as the total sales for a firm in a given year, divided by the total sales of the industry (SIC 1-digit) in that same year.}
\end{table}
